home *** CD-ROM | disk | FTP | other *** search
- From: evanlang@uss.lonestar.org (Evan Langlois)
- Subject: Diffs
- Date: Sat, 23 Jul 1994 11:41:47 -0500 (CDT)
-
- OK, I have finally worked all the bugs out of my file cookie cache
- routines. Well, all the serious ones :-) There are a few left where
- a cache entry is found when it shouldn't have been and sometimes the
- system will think a file is still in use when it isn't. These will
- have to be fixed gradually (the entire fcookie cache code is ifdef'd
- so you can compile without it).
-
- My changes are relative to Mint 1.10 h5 (with some other patches that
- didn't make it until h6). Now h7 is out. How do I apply the patches
- for h7 AND my own?? What diff command will create the proper diffs?
- Do I use diff3? And if so, how? I've gotten out of sync and I need
- help.
-
- Tests are real good. Preliminary tests bring 2 minutes 18 sec.s down
- to 1 minute 14 seconds. And another test that was an unbearable 7
- minutes and 6 seconds dropped to only 3 minutes and 4 seconds. So,
- access is down to half the time on average!
-
- Oh, I don't think the hacks that make VDI and AES allocates memory
- M_KEEP are useful. Under normal TOS, wouldn't this memory be attached
- to the process and free'd on exit?? I get no memory leaks at all
- changing the last of these so that memory allocation from the ROMs is
- no longer kept. I'm not sure if this breaks memory protection, but
- before I would lose a small amount of memory when some GEM programs
- quit, so perhaps maybe VDI and AES memory SHOULD be free'd on process
- exit ??
-
- As soon as someone tells how to go about making this diff, I'll post
- em and you can start fixing the last of the problems <grin>. Either
- that or I'll make the diffs and then apply them to the new kernel by
- hand. Most of the changes are large blocks and not one-liners, so
- I should be able to make the changes. Should this relative to h8?
-
- Thanks!
-
- ~.
-
-
-